Atmos. Chem. Phys., 12, 3045–3064, 2012
www.atmos-chem-phys.net/12/3045/2012/
doi:10.5194/acp-12-3045-2012
© Author(s) 2012. CC Attribution 3.0 License.
Atmospheric
Chemistry
and Physics
Evaluating WRF-Chem aerosol indirect effects in Southeast Pacific
marine stratocumulus during VOCALS-REx
P. E. Saide1 , S. N. Spak1 , G. R. Carmichael1 , M. A. Mena-Carrasco2 , Q. Yang3 , S. Howell4 , D. C. Leon5 , J. R. Snider5 ,
A. R. Bandy6 , J. L. Collett7 , K. B. Benedict7 , S. P. de Szoeke8 , L. N. Hawkins9 , G. Allen10 , I. Crawford10 , J. Crosier10 ,
and S. R. Springston11
1 Center
for Global and Regional Environmental Research (CGRER), University of Iowa, Iowa City, Iowa, USA
for Sustainability Research, Universidad Andrés Bello, Santiago, Chile
3 Pacific Northwest National Laboratory, Richland, WA, USA
4 Department of Oceanography, University of Hawaii at Manoa, Honolulu, USA
5 Department of Atmospheric Science, University of Wyoming, Laramie, Wyoming, USA
6 Department of Chemistry, Drexel University, Philadelphia, PA, USA
7 Colorado State University, Department of Atmospheric Science, Fort Collins, CO, USA
8 Oregon State University, College of Earth, Ocean, and Atmospheric Sciences, Corvallis, OR, USA
9 Harvey Mudd College, Department of Chemistry, Claremont, CA, USA
10 Centre for Atmospheric Science, University of Manchester, Manchester, M13 9PL, UK
11 Brookhaven National Laboratory, USA
2 Center
Correspondence to: P. E. Said (pablo-saide@uiowa.edu)
Received: 16 October 2011 – Published in Atmos. Chem. Phys. Discuss.: 4 November 2011
Revised: 9 February 2012 – Accepted: 8 March 2012 – Published: 29 March 2012
Abstract. We evaluate a regional-scale simulation with
the WRF-Chem model for the VAMOS (Variability of the
American Monsoon Systems) Ocean-Cloud-AtmosphereLand Study Regional Experiment (VOCALS-REx), which
sampled the Southeast Pacific’s persistent stratocumulus
deck. Evaluation of VOCALS-REx ship-based and three aircraft observations focuses on analyzing how aerosol loading
affects marine boundary layer (MBL) dynamics and cloud
microphysics. We compare local time series and campaignaveraged longitudinal gradients, and highlight differences in
model simulations with (W) and without (NW) wet deposition processes. The higher aerosol loadings in the NW case
produce considerable changes in MBL dynamics and cloud
microphysics, in accordance with the established conceptual
model of aerosol indirect effects. These include increase in
cloud albedo, increase in MBL and cloud heights, drizzle
suppression, increase in liquid water content, and increase in
cloud lifetime. Moreover, better statistical representation of
aerosol mass and number concentration improves model fidelity in reproducing observed spatial and temporal variability in cloud properties, including top and base height, droplet
concentration, water content, rain rate, optical depth (COD)
and liquid water path (LWP). Together, these help to quantify
confidence in WRF-Chem’s modeled aerosol-cloud interactions, especially in the activation parameterization, while
identifying structural and parametric uncertainties including:
irreversibility in rain wet removal; overestimation of marine
DMS and sea salt emissions, and accelerated aqueous sulfate
conversion. Our findings suggest that WRF-Chem simulates
marine cloud-aerosol interactions at a level sufficient for applications in forecasting weather and air quality and studying
aerosol climate forcing, and may do so with the reliability
required for policy analysis.
1
Introduction
Clouds play a major role in Earth’s radiative balance (Ramanathan et al., 1989; Cess et al., 1989). However, uncertainties in the processes that affect cloud optical properties
and modify this balance are still high (Solomon et al., 2007).
These processes are driven by the indirect climatic effects of
aerosols (Lohmann and Feichter, 2005), which can modify
cloud albedo (Twomey, 1991) and lifetime (Albrecht, 1989),
evaporate clouds (Graßl, 1979), change thermodynamics in
deep convective clouds (Andronache et al., 1999), increase
precipitation in ice clouds (Lohmann, 2002), and change the
surface energy budget (e.g., Liepert, 2002).
Low-level marine clouds have been shown to contribute
substantially to cloud radiative forcing (Ramanathan et al.,
1989). However, these clouds are not well represented by
contemporary models (Wyant et al., 2010). Previous work
Published by Copernicus Publications on behalf of the European Geosciences Union.
3046
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
has shown problems in the ability of global and regional
models to accurately represent marine stratocumulus clouds
(Vellore et al., 2007; Otkin et al., 2009; Wyant et al., 2010;
Abel et al., 2010), leading to difficulties in predicting cloud
cover on an operational basis (e.g., Shah et al., 2010). Some
problems are thought to be related to boundary layer schemes
generating insufficient vertical mixing, resulting in an unrealistically shallow cloud-topped boundary layer (Otkin and
Greenwald, 2008). By comparison, Large Eddy Simulation (LES) models have been shown to more effectively describe stratocumulus clouds and their transitions (e.g., Feingold et al., 1998; Khairoutdinov and Kogan, 2000; Berner
et al., 2011; Wang et al., 2010). Efforts have been made to
couple models at both scales (regional and LES), obtaining
accurate representation of stratocumulus (Zhu et al., 2010).
However, operational use of these coupled models for numerical weather prediction (NWP) or climate studies is not yet
feasible. Cloud data assimilation has been an alternative way
to improve clouds in NWP (e.g., Vellore et al. 2006; Errico
et al., 2007; Michel and Auligné, 2010).
Uncertainties in modeling aerosol indirect effects diminish our capability to generate reliable climate projections,
to evaluate policy questions and geo-engineering proposals, and to provide accurate weather and air quality predictions. Including indirect aerosol effects has been shown to
improve cloud representations in global models (Lohmann
and Lesins, 2002), and a range of approaches in modeling
them have been assessed (Ghan and Easter, 2006). On the
regional scale, including aerosol indirect effects tends to impact clouds optical properties (Chapman et al., 2009) and
precipitation (Ntelekos et al., 2009), and often produces better cloud representation by optical properties, dynamics and
microphysics (Gustafson et al., 2007; Q. Yang et al., 2011).
The LES scale has been able to show the effect on cloud
structure by different cloud condensation nuclei (CCN) loadings, and effectively simulate the dynamics of open cells,
“pockets of open cells,” and closed cell marine clouds (Wang
and Feingold, 2009a, b).
Intensive measurement campaigns provide a wealth of observations that present the opportunity to evaluate models
and to identify, quantify, and hopefully reduce these uncertainties. The VAMOS Ocean-Cloud-Atmosphere-Land
Study Regional Experiment (VOCALS-REx, Wood et al.,
2011) was an international field program designed to make
observations of poorly understood but critical components of
the coupled climate system of the southeast Pacific on the
coast of Chile and Peru. Reactive gas and aerosol observations show a marked longitudinal gradient from elevated
values close to shore due to polluted conditions to cleaner
remote conditions (Allen et al., 2011), while cloud properties correlate to some extent with this gradient (Bretherton
et al., 2010). Model evaluation studies emerging from the
campaign have identified difficulties in accurately representing MBL and stratocumulus clouds (Abel et al., 2010; Sun
et al., 2010; Andrejczuk et al., 2011) without considering
Atmos. Chem. Phys., 12, 3045–3064, 2012
aerosol feedbacks to meteorology. Q. Yang et al. (2011)
present a comprehensive evaluation of the WRF-Chem system on a regional scale highlighting the effects of aerosol
feedbacks, showing that the inclusion of aerosol-cloud interactions typically improve model performance in simulating
cloud properties.
In this work, we build upon previous regional simulations
including aerosol feedbacks using the WRF-Chem model.
Several modeling studies have performed sensitivity analyses of the effects of aerosol loading on cloud properties
(e.g., Chen et al., 2011). Starting from a base configuration,
we find another configuration that better represents aerosol
mass and number concentrations, and then analyze the impacts of these different aerosol loading on MBL dynamics
and cloud microphysics, and compare them to observations
and to the canonical conceptual model of warm cloud indirect effects. We perform an extensive evaluation of different
aspects of the model representation, and identify areas for
improvements and remaining problems.
2
Methods
For the purposes of defining representative spatial zones
characterized by broadly internally similar thermodynamic
aerosol and composition regimes (when averaged over the
length of the VOCALS-REx campaign) we choose to use the
three areas as defined by Bretherton et al. (2010) and Allen
et al. (2011). These are the “coastal zone” (or “off shore”,
east of 75◦ W), and the “remote zone” (west of 80◦ W), with
the two regions separated by a “transition zone” near the
78◦ W meridian (75◦ W–80◦ W).
2.1
WRF-Chem model configuration
The WRF-Chem model simulates meteorology and atmospheric constituents, as well as their interactions (Skamarock
et al., 2008; Grell et al., 2005). We configured WRFChem v3.3 with a combination of model structures, parametric choices, and input data to best represent marine stratocumulus conditions, atmospheric chemistry, and secondary
aerosols, with the goal of future use in meteorological and air
quality forecasting.
A 12×12 km2 horizontal resolution domain is employed,
covering 91◦ W–65◦ W longitude and 40◦ S–12◦ S latitude.
This choice attempts to optimize between spatial resolution, critical for representing cloud dynamics, and complete
coverage of the VOCALS region of the Southeast Pacific
within the limitations of computing time. The domain accounts for most major Chilean and Peruvian anthropogenic
sources shown in airmass trajectories (Chand et al., 2010) to
impact VOCALS-REx observations, including the Andean
cordillera, and covers the entire VOCALS-REx experimental domain (to ∼85◦ W), with a margin towards the west
and north to avoid excessive boundary condition influence
on meteorology and atmospheric composition. Following
recommendations by Wang et al. (2011), a 75-level vertical
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
resolution was chosen to reduce MBL and cloud height underestimation. The first few levels are as in Saide et al. (2011)
with ∼10 m thickness, and the average vertical layer spacing
between 60 m and 3 km is ∼60 m. In preliminary testing,
this resolution produced accurate MBL and cloud heights for
all longitudes, which were ∼100–300 m greater than the 39level resolution used in Saide et al. (2011).
Model structure was configured to combine modules included in contemporary WRF-Chem public release code that
best represent known aerosol, cloud, and MBL processes and
their couplings. Wherever possible, the most complete representations of complex physical and chemical processes were
chosen. This application requires a boundary layer closure
scheme that can make use of (and maintain numerical stability at) high vertical resolution, and can accurately represent
the diurnal evolution of the MBL at low wind speeds. MellorYamada type schemes have generally exhibited good cloud
representation under these conditions (Otkin and Greenwald,
2008; Zhu et al., 2010; Rahn and Garreaud, 2010). The
MYNN level 2.5 scheme (Nakanishi and Niino, 2004) was
chosen since it performed well in prior applications at this
resolution over Chile (Saide et al., 2011). The Lin microphysics scheme (Chapman et al., 2009) and Goddard short
wave radiation (Chou et al., 1998; Fast et al., 2006) were
chosen to support aerosol direct, indirect, and semi-direct
feedbacks to meteorology. Activation of aerosols from the
interstitial to the cloudborne “attachment state” (Ghan and
Easter, 2006) is based on a maximum supersaturation determined from a Gaussian spectrum of updraft velocities and
the internally mixed aerosol properties within each aerosol
size bin (Abdul-Razzak and Ghan, 2002). The updraft velocity distribution is centered in the model vertical wind
component plus the subgrid vertical velocity diagnosed from
vertical diffusivity. No cumulus scheme was used following the recommendation of Q. Yang et al. (2011). The
RRTM longwave radiation scheme (Mlawer et al., 1997) was
used. Gases and aerosols were simulated using the CBMZ
gas-phase chemical mechanism (Zaveri et al., 1999; Fast
et al., 2006) with dimethyl sulfide (DMS) reactions coupled to the 8-bin sectional MOSAIC (Zaveri et al., 2008)
aerosol module. Seawater DMS concentration was set to
2.8 nM, following the VOCA Modeling Experiment Specification (http://www.atmos.washington.edu/∼mwyant/vocals/
model/VOCA Model Spec.htm) and in agreement with measurements during VOCALS-REx (Hind et al., 2011). DMS
is transferred to the air using sea-air exchange as in Liss and
Merlivat (1986).
We chose emissions and chemical boundary conditions to
best resolve spatial and temporal variability in aerosols and
their precursors, taking into account a complete range of natural and anthropogenic emissions sources. Continental emissions of biogenic trace gases (e.g., isoprene) were predicted
hourly by the MEGAN algorithm (Guenther et al., 2006),
and daily biomass burning locations and fuel loadings were
obtained from FIRMS MODIS fire detections (Davies et al.,
www.atmos-chem-phys.net/12/3045/2012/
3047
2009) and modeled hourly using WRF-Chem’s plume rise
model (Freitas et al., 2006, 2007). Volcanic and anthropogenic emissions, including point and area sources, were
taken from the VOCA inventory described in detail by MenaCarrasco et al. (2012). Table 1 shows a summary of the
sources of information for the emission inventories compiled for this research. In cases where particulate matter
(PM) was not speciated, 10 %, 30 % and 70 % were associated to elemental carbon, organic carbon and crustal aerosol,
respectively. Chemical boundary conditions are obtained
from 6-hourly MOZART global simulations (Emmons et al.,
2010). MOZART fields were found to overestimate nearshore concentrations, so the model was started from clean
initial conditions and spun up for 6 days to avoid biasing
results. MOZART sulfur dioxide (SO2 ) boundary conditions in the free troposphere (FT) were found to be underestimated, so a global minimum background level of 30 ppt
and a 50 ppt minimum for heights over 3.5 km were set, in
agreement with flight profile measurements in the remote region (Allen et al., 2011; Kazil et al., 2011). Sea salt aerosol
emissions were modeled following Gong et al. (1997), but
resultant concentrations from the default scheme were found
to substantially overestimate ship-based measurements from
the NOAA RV Ronald H. Brown (Ron Brown). In order to
avoid misleading indirect effects due to these biases, submicron emissions were reduced by a factor of 10 and supermicron emissions were reduced by a factor of 2, in line with
campaign-averaged observations from the Ron Brown. Default WRF-Chem sea salt emissions do not consider sulfate
coming from seawater, speciating sea salt as Na and Cl only.
Wind-blown dust was not modeled, due to known high biases
in WRF-Chem’s online wind-blown dust emissions, concentrations, and resultant aerosol optical depth over land, and
poor model representation of Andean dust composition. No
organic sea emissions were considered in this study, as there
was little to no evidence of these submicrometer particles
during the campaign (Shank et al., 2012). Also, no secondary
organic aerosols (SOA) were modeled as the fraction of SOA
to total organic aerosol is thought to be low in this region
(∼10 %, Kanakidou et al., 2005). New particle formation is
modeled by the Wexler et al. (1994) scheme.
WRF-Chem simulations covered the entire VOCALS-REx
campaign period, 15 October–16 November 2008, along
with the extra 15 days that Ron Brown stayed in the domain (16–30 November). The model was run with an initial “chemical” spin-up period of 6 days with meteorology
re-initialized from analyses at the middle of the modeling
period using the previous chemical state. We found that 3–
4 days of spin up are necessary to overcome the underestimated MBL height present in the National Centers for Environmental Prediction (NCEP) Final Analysis (FNL) from
GFS (Sun et al., 2010), as only increasing vertical resolution
is insufficient for debiasing the offshore MBL (Andrejczuk
et al., 2011).
Atmos. Chem. Phys., 12, 3045–3064, 2012
3048
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Table 1. Summary of emissions inventories configurations. Areas are divided in metropolitan region, which is located on Central Chile, the
rest of Chile and the rest of the domain.
Geographical area
Sector
Inventory name
Species
Metropolitan
region
Mobile sources
SECTRA
Metropolitan
region
Residential sources
Metropolitan
region
Point sources
Rest of Chile
Power plant
emissions
Rest of Chile
Smelter emissions
Rest of Chile
Mobile sources
Rest of domain
Total anthropogenic excluding power
and smelting
Total anthropogenic
Chilean Ministry of
environment
DICTUC (2007)
Chilean Ministry of
environment
DICTUC (2007)
Chilean power plant
emissions standard
KAS (2009)
Chilean air quality
standards revision
Mena-Carrasco (2010)
SECTRA regional
Corvalán et al. (2005)
EDGAR 3.2
Olivier et al. (1994)
PM10 , PM2.5 , CO,
NOx , SOx , VOCs,
NH3
PM10 , PM2.5 , CO,
NOx , SOx , VOC,
NH3
PM10 , PM2.5 , CO,
NOx , SOx , VOC,
NH3
PM10 , PM2.5 , CO,
NOx , SOx , VOC,
NH3
PM10 , PM2.5 , CO,
NOx , SOx , VOC,
NH3
PM10 , PM2.5 , CO,
NOX, VOC
PM10 , PM2.5 , CO,
NOx , SOx , VOC,
NH3
Black carbon and organic carbon
Rest of domain
Bond et al. (2004)
We found that aerosol wet deposition has a large influence over the modeling results. In the present version of
WRF-Chem, in- and below-cloud wet removal of gases and
aerosols in CBMZ-MOSAIC are modeled following Easter
et al. (2004). This mechanism assumes that the removal
processes are irreversible, and does not consider aerosol resuspension due to rain evaporation. This becomes an important issue for the Southeast Pacific during Austral spring,
since most of the drizzle observed during VOCALS-REx
evaporated before reaching the surface (Bretherton et al.,
2010), leading to a great contrast between cloud base and
surface rain rates. Thus, irreversible removal of aerosol by
rain might create an unrealistically strong sink, which is supported by previous modeling results (Q. Yang et al., 2011).
Kazil et al. (2011) implemented wet removal considering rain
evaporation, but for a different modal aerosol approach and
in the context of LES simulations. To assess the importance
of modeled wet removal processes, we performed simulations where wet deposition was excluded, which results in
higher aerosol loadings. This represents an upper limit to
below cloud aerosol, and reflects the fact that low rain rates
were observed at the sea surface (0.01 mm h−1 on average)
during the VOCALS campaign (M. Yang et al., 2011), indicating that most rain evaporated before reaching the surface
(Bretherton et al., 2010), suggesting nearly zero wet deposition. Thus, by turning off wet deposition the unrealistic sink
of aerosol mass generated by not considering resuspension is
removed. However, the effects in terms of number concentration are uncertain due to complex interactions: one droplet
Atmos. Chem. Phys., 12, 3045–3064, 2012
Base year
2010
2005
2005
2009
2010
2005
2005
2005
can collect thousands of particles by collision-coalescence
but, as some have observed (Mitra et al., 1992; Feingold
et al., 1996), only one aerosol is released after evaporation. Since rain rates increase as aerosol number concentration decrease, a cloud-scavenged ultra-clean layer can be
generated which can lead to conditions of particle nucleation (Kazil et al., 2011) potentially recovering the number
of particles lost before. Without an aerosol module that includes reversible wet deposition, and for the sake of studying the sensitivity to different aerosol loads, both simulations
were conducted for the whole period. The simulation with
wet deposition turned on is hereafter referred as the base
run or “W”, while the simulation without wet deposition
is called “NW”. Since W represents large aerosol removal,
while NW no aerosol removal, we hypothesize that a model
with a correct wet deposition scheme should be bounded by
these two states.
2.2
Observations
The observations used for comparison are provided by the
VOCALS-REx airborne and marine platforms. Carbon
monoxide (CO), Ozone (O3 ), SO2 , DMS gases; and sulfate
(SO4 ), nitrate (NO3 ), ammonium (NH4 ) and organic carbon
(OC) interstitial aerosol (from AMS: non-refractory non-sea
salt) observations collected by the NSF C-130, DoE G-1,
FAAM BAe-146 aircrafts and Ron Brown are thoroughly
described by Allen et al. (2011), Kleinman et al. (2012),
M. Yang et al. (2011), and Hawkins et al. (2010), while
C-130 cloud water composition measurement methods can
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
be found in Benedict et al. (2012).
The University
of Wyoming 94 GHz cloud radar (WCR) aboard C-130
provided radar reflectivities, which were then corrected
(Bretherton et al., 2010) and converted to rainfall estimates using the Z-R relationship described in Comstock
et al. (2004). This presents results consistent with Particle
Measuring Systems (PMS) Two Dimensional Cloud Probe
(2D-C) probe rainfall estimates during VOCALS (Bretherton et al., 2010). The WCR, along with an upward-pointing
lidar (WCL) provided cloud top and base height estimates
from the C-130 (Bretherton et al., 2010). Cloud top and base
heights from Ron Brown were estimated using a millimeterwave cloud radar (MMCR) and a Vaisala CL31 ceilometer, respectively (de Szoeke et al., 2010). Capping inversion
height (CIH) was estimated as the height at which the temperature was a minimum, provided the relative humidity was
at least 45 % (Jones et al., 2011) in both Ron Brown soundings (Wood et al., 2011) and aircrafts vertical profiles. C-130
Gerber PVM-100 Probe cloud water content, PMS Cloud
Droplet Probe (CDP) and Forward Scattering Spectrometer
Probe (FSSP-100) cloud droplet number concentration, and
PMS Passive Cavity Aerosol Spectrometer Probe (PCASP)
accumulation mode aerosol number concentration observations used are described in Kazil et al. (2011) and Bretherton et al. (2010). BAe-146 cloud and accumulation mode
aerosol measurements (Allen et al., 2011) were performed
with similar instruments as in C-130 (Droplet Measurement
Technologies (DMT) CDP-100, PCASP) while G-1 (Kleinman et al., 2012) used a DMT Cloud and Aerosol Sampling
(CAS) probe and a PCASP, respectively. An intercomparison of the cloud microphysics probes fitted to BAe-146 and
C-130 was performed on 31 October 2008 and 4 November 2008. The aircraft performed straight and level runs (of
the order of 10’s of km in length) through the same region
of cloud approximately 5 min apart, finding that the number concentration, LWC and size distributions were similar within calibration and systematic error. However, G1 cloud microphysics measurements showed inconsistencies
compared to other probes used (Kleinman et al., 2012) probably due to shattering of drizzle on CAS inlet (McFarquhar
et al., 2007). On the Ron Brown, total number of particles
over 13 nm was measured with a TSI 3010 Condensation Particle Counter. Cloud optical depth (COD), cloud liquid water
path (LWP) and cloud effective radius were obtained from
MODIS-Aqua retrievals.
2.3
Performance statistics
We present box and whisker plots of longitudinal profiles at
20◦ S (e.g., Fig. 1) in order to assess model performance in
a consistent manner across trace gas, aerosol, and cloud properties, and to focus evaluation on the longitudinal gradients
identified in VOCALS-REx observations as the most important characteristics of aerosol and low cloud regimes in the
Southeast Pacific (Allen et al., 2011; Bretherton et al., 2010).
www.atmos-chem-phys.net/12/3045/2012/
3049
Table 2. Measure of model performance using data obtained from
a box and whisker plot. A “match” is defined as a model (observation) median or mean being in between the two percentiles of the
observation (model) distribution. Matches ranges from 4 (perfect
match) to 0 (no match). An “overlap” is defined as when modeled and observed inner quartiles (boxes) or inner deciles (whiskers)
overlap. Overlaps ranges from 2 to 0. The final performance is
assigned as the best of the three criteria.
Excellent
Very good
Good
Fair
Poor
Matches in
the 25th–75th
percentile
Matches in
the 10th–90th
percentile
Overlaps
4 or 3
2 or 1
–
–
–
–
–
4 or 3
1 or 2
–
–
–
2
1
0
Here, we introduce a measure of accuracy deduced from
these plots, hereafter referred to as the Box and Whiskers
(BoW) metric and summarized in Table 2. We define a
“match” as a model (observation) median or mean falling
between the two prescribed percentiles of the observation
(model) distribution. The first criterion (Table 2, column
1) uses the 25th and 75th percentiles (boxes in Fig. 1) and
the second criterion (Table 2, column 2) uses the 10th and
90th percentiles (whiskers in Fig. 1). The third criterion (Table 2, column 3) is based on overlapping of the 25th and
75th (box overlapping) and 10th and 90th (whisker overlapping) percentile distributions. The net level of accuracy is
determined from the sum of these three criteria, which is
converted into a qualitative category: excellent, very good,
good, fair and poor. Some of the advantages of the BoW
method are that: it is independent of the variable being assessed; no absolute threshold of accuracy is specified for any
variable; it is based on basic statistical parameters; it can easily be read from a box and whisker plot; and it transforms
a quantitative measure of accuracy into a qualitative description. However, some issues could be encountered when the
distributions are strongly skewed, as the mean could be found
outside the inner quartile.
Statistics for cloud microphysics and aerosol number concentration were computed for aircrafts profile means instead
of local point to point comparisons, since observed clouds
could be in different levels than model values, generating
mismatches with modeled and observed clouds both present,
but at different levels. For estimating the modeled 117 nm
to 3 µm PCASP aerosol number concentration, values from
bins 3 to 7 (156 nm to 2.5 µm) are integrated along with 42 %
of the second bin (78 to 156 nm), which corresponds to the
fraction over 117 nm using the logarithmic diameter.
Aircraft modeled and observed gases, aerosol mass, cloud
heights and rain statistics were computed for one minute
Atmos. Chem. Phys., 12, 3045–3064, 2012
3050
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
(a)
(b)
(c)
(d)
(e)
Fig. 1. Observed and modeled statistic for selected gaseous and aerosol species gridded into 2.5 degree longitudinal zones in between 22◦ S
and 18◦ S. For each zone, centre solid (dashed) lines indicate the median (mean), boxes indicate upper and lower quartiles with upper and
lower decile whiskers. The sampling time in decimal hours in each longitude bin is indicated at the top. Left column and right column are
for marine boundary layer (MBL) and free troposphere (FT), respectively.
average values, while Ron Brown statistics were computed
for ten minute intervals. For the case of rain statistics, model
results are not filtered for missing observations and vice
versa as information on rain frequency can be extracted from
the total sampling time in each longitude bin on the top of
the box and whisker plots (e.g., Fig. 1). This creates minor inconsistencies; mainly in the 100 m height estimates,
which was not measured in sub-cloud flight legs (Bretherton
et al., 2010).
3
Results and discussion
We first focus on evaluating atmospheric concentrations of
selected gases and aerosols for the base and NW simulations.
Then, model performance is assessed for MBL dynamics and
Atmos. Chem. Phys., 12, 3045–3064, 2012
cloud microphysics. Finally, spatial and temporal variability
in chemical transport and cloud effects are investigated on an
episodic basis.
3.1 Trace gas and aerosol evaluation
Figure 1 shows aircrafts flight statistics for gaseous and
aerosol concentration for selected species for the MBL and
free troposphere (FT). For this plot, MBL concentrations
were considered for heights lower than 1200 m or below
cloud, and FT concentrations for heights in between 1700
and 3200 m in order to avoid cloud contamination (Allen
et al., 2011). Trace gas data quality was assessed by intercomparison between aircrafts, obtaining 1.5 and 4 ppb as the
uncertainties for CO and O3 , respectively (Allen et al., 2011).
As seen in Fig. 1, MBL CO is overestimated across the entire
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
modeled longitudinal range with the exception of the close
to the shore bin where highly polluted plumes were detected
by the G-1. Measurement uncertainty is well below these
differences pointing to a model bias. Neglecting these nonresolved plumes, close to shore overestimation is probably
due to a lower MBL than observed (see Sect. 3.2), while remote zone issues are likely due to overestimation of the MBL
CO MOZART boundary conditions, as these air masses often had no contact with the continent for a long time period
(Allen et al., 2011). However, we cannot rule out the possibility that a combination of overestimates in Central Chile
anthropogenic emissions (e.g., Jorquera and Castro, 2010;
Saide et al., 2009) and too much entrainment in the model
could generate MBL concentrations similar to FT concentrations. The latter case is less likely, as it would have similarly
affected O3 . Remote FT CO shows very good to excellent
performances driven mostly by MOZART boundary conditions over the east-central Pacific. Even though MBL CO
shows poor to fair performance in BoW metrics, differences
are no more than 15 ppb, and the observed longitudinal trend
(decreasing towards remote zone) and spread (<10 pbb) are
often well simulated, indicating that transport in the MBL is
resolved. The base and NW models show very small differences, attributable to changes in entrainment and MBL
heights (see Sect. 3.2).
Figure 1b shows O3 was well simulated for both the MBL
and FT, with very good to excellent BoW metrics, and with
similar spread. There is no clear longitudinal trend in either
the model or the observations. An important point is that,
as mentioned in Allen et al. (2011), the model resolves the
∼30 ppb difference between MBL and FT and also the higher
variability in the FT. The lower O3 in the MBL is the result
of chemical destruction during the day, transport from FT
during the night due to entrainment (M. Yang et al., 2011),
and lower photolysis rates and temperatures under the cloud
deck. Due to the ability of the model to correctly maintain
the MBL to FT O3 difference, we surmise that entrainment
is simulated effectively. Hydroxyl radical (OH) in VOCALS
MBL was estimated by Yang et al. (2009) from the DMS
budget and found to have maximum diurnal values of 3–5 ×
106 molecules cm−3 . WRF-Chem showed OH peaks in the
lower end of this range, at ∼2.5–3.35×106 molecules cm−3 .
Statistics for gas and aerosol components of the sulfur cycle are shown in Fig. 1c–e. The C-130 measured FT DMS
(Fig. 1c, right panel) was usually below the detection limit
(5 ppt), as in the model. The spikes in DMS for the 75th
to 90th percentile show times where the cloud top heights
were >1700 m, which is better captured by NW as explained
later. In general, MBL DMS has a high bias, with poor to
fair BoW scores. As discussed by Q. Yang et al. (2011), this
is likely related to an overestimation of DMS emissions due
to overestimation of the modeled DMS ocean : atmosphere
transfer velocity. Similarly to CO, and despite the emission
bias, the modeled longitudinal trend is captured very well
by the model. Ron Brown atmospheric DMS measurements
www.atmos-chem-phys.net/12/3045/2012/
3051
Table 3. Observed and modeled cloud chemistry statistics. Values
where the observations (model) were inexistent were removed from
the model (observations) statistics.
Observation
Mean Std dev
pH
H2 O2 (µM)
TOC (µg C L−1 )
Cl− (µN)
NO−
3 (µN)
SO2−
4 (µN)
Na+ (µN)
NH+
4 (µN)
Ca2+ (µN)
4.94
132.20
2028.9
855.3
72.77
298.35
1204.6
90.65
128.60
0.91
130.94
571.0
1205.0
120.39
465.94
2008.6
163.89
203.39
NW model
Mean Std dev
4.12
153.44
1624.6
1220.2
28.41
133.42
1253.3
2.51
0.00
0.43
73.47
1953.0
1450.7
32.22
152.71
1486.8
2.88
0.00
showed higher values, in better agreement with the model
but still lower. FT SO2 is also skillfully simulated, presenting mostly excellent BoW metrics and follows the observed
longitudinal trend. FT Remote zone SO2 is mostly affected
by boundary conditions, showing the importance of setting
lower thresholds for influx from MOZART (see Sect. 2). In
the MBL, the model usually exhibits very good performance,
but cannot maintain the ∼20 ppt lower threshold observed.
When looking at the modeled SO2 diurnal cycle, the higher
values are obtained after DMS photochemical destruction,
but rapidly decay to values as low as 1 ppt due to cloud processing and conversion to SO4 aerosol. Modeled conversion
appears to occur at a higher rate than observed, which is investigated later in more detail. Finally, Fig. 1e shows sulfate mass concentrations statistics. MBL SO4 clearly shows
the impact of wet deposition as no remote zone W and NW
model distributions overlap. The observations are typically
closer to the NW results, and sometimes between the two
simulations. As most of the rain evaporated before reaching
the sea surface, we find that the NW results are more realistic, and that any overestimation could be due to the combination of high DMS emissions rates and high SO2 to SO4 cloud
conversion yields. In the FT, both configurations show similar results with very good to excellent BoW performances.
Other species play much smaller roles in aerosol composition. The model significantly underestimates NH4 (possibly
due to emissions, which are poorly constrained in the region),
estimates of NO3 are below detection limits as observed, and
accurately predicts organic carbon (not shown). Most modeled NO3 is found in the coarser sectional bins, as it is displaced by SO4 in fine aerosol bins due to the low NH3 concentrations (Seinfeld and Pandis, 2006), and therefore rarely
appears in aircrafts AMS observations, where aerodynamic
diameter is capped at 500–700 nm. Coarser aerosol is dominated by sea-salt, where NO3 displaces Cl creating a chloride
deficit (Q. Yang et al., 2011).
Atmos. Chem. Phys., 12, 3045–3064, 2012
3052
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
(a) C-130 Observations
(b) WRF-Chem NW all bins
(c) WRF-Chem NW bins 1-3
Fig. 2. Pie charts for modeled ionic species for C-130 observations representing cloud composition (a) and the no wet deposition model
(NW) using collection of wet aerosol along the flight track for all bins (b) and for bin 1, 2 and 3 (40 nm to 300 nm aerosol diameter) only (c).
Units are in µN.
In order to further explore SO2 to sulfate conversion processes, we compare cloud chemistry observations to the NW
model (Fig. 2 and Table 3), as both models show similar
cloud aerosol composition. Consistent with observations
(Benedict et al., 2012), model results show that bulk (summing all sizes) cloud drop ion concentrations are dominated
by sea salt, followed by sulfate (Fig. 2a, b). Bulk sulfate concentrations are underestimated, since sulfate coming from
seawater is not modeled. As shown by Table 3, in general
the model does a good job representing the mean and variability of the ion concentrations. The most notable problems
are Ca+2 , which is very low since no dust was modeled, and
NH4 , which is underestimated as it was in the AMS interstitial aerosol evaluation. Model pH shows the same tendency
as observations, increasing towards the remote region as sulfate aerosol is more abundant close to shore (not shown).
However, model pH is always under 5, while values up to
7 were observed, leading to underprediction in mean pH (Table 3). We found that the bulk model is extremely sensitive to
chloride concentrations, as a decrease in only 5 % in Cl− (as
in observations) will increase average pH by 1 and increase
single values up to 2.5 pH units. This is important, as WRFChem uses a bulk cloud chemistry scheme (Chapman et al.,
2009) and small variations in Cl− (thus in pH) can generate
a shift in the dominant mechanism of SO2 to sulfate conversion, from the roughly pH independent H2 O2 reaction (Martin and Damschen, 1981) to the O3 reaction which increases
in rate with pH (Hoffmann and Calvert, 1985), resulting in
a speeding up of the SO2 to sulfate conversion and even further reductions in SO2 concentrations. However, as pointed
out by M. Yang et al. (2011), most droplets nucleate from
sulfate particles, so their pH will be acidic and dominated by
the hydrogen peroxide reaction. This behavior for the majority of droplets is seen in modeled cloud water aerosol in the
bins that dominate nucleation (Fig. 2c). We also see very low
sea salt influence, as Na+ percentage is low and Cl− is diffused into the droplet from HCl gas rather than entering the
droplet as sea salt. All this implies that there is clear need
Atmos. Chem. Phys., 12, 3045–3064, 2012
for sized-resolved cloud chemistry (e.g., Fahey and Pandis,
2001), and that aqueous chemistry should be considered for
nucleation and accumulation modes only (Kazil et al., 2011).
Measurements considering the nature of the cloud condensation nuclei (CCN) composition and size should also be performed (Bator and Collett, 1997). Analyzing the H2 O2 pathway, H2 O2 concentrations are slightly overestimated by the
model (Table 3), which cannot explain the SO2 gap between
model and observations. M. Yang et al. (2011) found that
to close the SO2 budget, the Martin and Damschen (1981)
H2 O2 rate expression yielded best results, while other reaction rates were too fast to reach mass balance. We compared
these rates to the McArdle and Hoffmann (1983) rates implemented in WRF-Chem, reaching the same answer, which
could explain the difference in SO2 . Another factor that influences increased SO2 depletion is the consistent overestimation of cloud fraction, as WRF-Chem NW shows average cloud fractions of 86 % on the Ron Brown track, while
the MMCR on board of Ron Brown (M. Yang et al., 2011)
showed values of 67 %.
3.2
MBL and marine Stratocumulus dynamics
Box and whisker plots for cloud base and cloud top for the
VOCALS-REx period are shown in Fig. 3a, b. The WRFChem NW model shows up to 200 m higher mean and median cloud top and base heights than the base model, bringing it closer to observations. The largest differences are
found in the remote zone. The higher accumulation mode
aerosol load obtained by NW allows for a less broken cloud
deck with smaller droplet radius and less precipitation (see
Sect. 3.3), which affects the MBL energy budget by decreasing average downward shortwave radiation (SW), upward
surface heat flux and top of the atmosphere (TOA) outgoing longwave radiation (LW) by 50–60 W m−2 , ∼3 W m−2
and 1–3 W m−2 , respectively. Precipitation, SW absorption
and LW cooling of clouds are the main drivers of entrainment
in a cloud topped boundary layer, which in turn determines
its cloud height. As shown by Pincus and Baker (1994),
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
(a)
(b)
g
(c)
Fig. 3. Box and whisker plots for different variables derived from
aircrafts measurements as in Fig. 1. (a) and (b): cloud top and bottom from U. of Wyoming radar (WCR) and lidar (WCL), respectively. (c) Decoupling index, the horizontal dashed line indicates
the 0.5 g kg−1 decoupling threshold (see explanation on the text).
The numbers above each zone represent sampling time in decimal
hours for (a) and (b), and number of profiles for (c).
when number concentration of droplets increase, precipitation decreases, which increases entrainment. However, this
also generates extra cloud water that produces thicker clouds
that absorb more shortwave radiation (lower model downward SW), heating the layer and decreasing entrainment.
Also, when clouds rise, cloud top temperature tends to decrease, decreasing LW cooling (model TOA outgoing LW
decrease) and thus reducing entrainment. An overall increase
in entrainment is achieved which cause cloud heights to rise
(Pincus and Baker, 1994), in agreement with our results.
As aerosol loads increase for NW, the direct and semidirect effects are also expected to change. However, semidirect effects should not play an important role as BC observations (Shank et al., 2012) and model results show very
low concentrations. A simulation where the aerosol radiation feedbacks were turned off using the NW configuration
shows small differences for cloud top pressure (<±1 %),
cloud fraction (<±10 %) and water content (<±10 %), implying that indirect effects dominate under clean conditions
like those observed during VOCALS-REx, where aerosol radiative effects become more important under heavily polluted
conditions (Koren et al., 2008).
As seen in Fig. 3a, b, close to shore both simulations have
large cloud height negative biases (fair to good BoW), as the
coarse resolution is unable to resolve the steep topography
and land-sea transition (Wang et al., 2011). In the remote
www.atmos-chem-phys.net/12/3045/2012/
3053
zone the NW heights captured well the observations (mostly
excellent BoW classifications) while the base model underestimated the heights (fair to excelent BoW scores). The NW
model also better represents observed temperature and water
vapor profiles in the remote zone both from aircraft profiles
and ship-based soundings (not shown), as the typical MBL
structure approaches the observed vertical profile. Even so,
there are still some periods where the model does not simulate the very high cloud heights observed in the remote zone,
as depicted by the 95th extremes of the observed distribution and as seen in the Ron Brown time series in Fig. 4,
which are responsible for the lower model means. However,
these periods of poor performance appear episodic, and there
are periods where WRF-Chem NW does reach the observed
heights (e.g., RF03 and RF05 on Fig. 4). Episodic underestimation of cloud heights is thought to come from meteorological boundary condition issues, as the model is unable
to represent the high clouds condition occurring over several
days (e.g., 19–23 November). The model shows good agreement with observations for a range of very different conditions: a 20S/POC drift flight (RF02) with very thin clouds,
two flights to 85 W (RF03 and RF05) with different longitudinal cloud trends, and a coastal pollution survey flight
(RF12) capturing the latitudinal gradient in cloud height.
In order to explore the model representation of MBL dynamics in more detail, a decoupling measure was computed.
Jones et al. (2011) showed that an effective decoupling indicator can be calculated as the difference in total water mixing ratio (qt , water vapor plus cloud water) between two levels: 25 % and 75 % of the capping inversion height (CIH),
considering a value below (above) 0.5 g kg−1 as a coupled
(decoupled) MBL. Observed decoupling index and capping
inversion height were obtained here from aircrafts vertical
profiles following the method of Jones et al. (2011), and
modeled values were obtained mapping the profiles and computing a modeled CIH and decoupling index. Figure 3c
shows longitudinal statistics for all aircrafts flights. Both
simulations represent several basic aspects of the observed
decoupling. The modeled decoupling index is accurately
predicted everywhere (very good to excellent BoW performance) but in the transition zone, where performance is
lower but still good. On average, areas west of 78◦ W are
decoupled while areas east of 78◦ W are coupled both in the
observations and the model, and better represented by the
NW simulation. The spread of the decoupling index on each
zone is also well simulated, with noticeable higher spread
west to 78◦ W, as these zones alternate in between coupled
and decoupled MBLs. Observations show a sharp longitudinal transition from coupled to decoupled MBLs, which is
also represented by the model.
3.3
Cloud microphysics
Figure 5 shows statistics for cloud properties and for aerosol
number concentration. Model cloud water representation is
Atmos. Chem. Phys., 12, 3045–3064, 2012
3054
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Fig. 4. Observed and NW model cloud bottom, cloud top and capping inversion height (CIH) time series from Ron Brown (top) and four
C-130 flights (bottom). Shaded areas represent night periods.
very good to excellent (in BoW metrics) for both models
(Fig. 5a), but the NW model typically shows higher amounts
of cloud water than the base model as clouds are more permanent and thicker (consistent with the Twomey effect found
by Albrecht, 1989). A clear difference is seen when analyzing number of droplets (Fig. 5b) where the increase in cloud
albedo is more evident (Twomey, 1974) and modeled inner
quartiles do not overlap over the remote region. This is a result of the difference in sub-cloud aerosol number concentration (Fig. 5c), where not even the modeled deciles overlap.
Comparing observed and model droplet number and aerosol
number concentration in the remote zone, the NW model
presents excellent results while the base model is biased low
(poor to good BoW scores), showing vast improvements in
cloud microphysics by increasing the sub-cloud aerosol to
near observed levels. We found consistency in the results, as
when aerosol loads are relatively close to observations, the
number of droplets also becomes closer to the observations.
We thus conclude that the activation routine in WRF-Chem is
consistent and reliable. The inability of NW to represent the
lower end and spread of the cloud droplet and aerosol number distributions can be related to not considering wet deposition, as aerosol number is expected to decrease for intense
sub-cloud rain events since a single droplet can collect a large
amount of particles and release just one when evaporation
occurs (Kazil et al., 2011). Close to shore overestimation
of droplet number concentration by both simulations may be
explained by the slight overestimation of aerosol number and
also by the fact that the model finds that the aerosol number concentration in the 1st bin (40 to 78 nm in diameter) is
an important contributor to activated particles. The latter is
Atmos. Chem. Phys., 12, 3045–3064, 2012
not captured by the PCASP aerosol number concentrations,
as it only measures aerosol diameters over 117 nm. Free
troposphere aerosol number concentration (Fig. 5d) follows
the same trend as in the MBL, with good to excellent BoW
accuracy and few differences between the two simulations.
Rain estimates extracted from radar reflectivities and
model statistics are shown in Fig. 6. Focusing on the cloud
top (a) and cloud base (b) plots, it can be seen that the model
captures in-cloud rain stratification, showing lower mean and
median values for the cloud top. WRF-Chem also represents
the longitudinal gradient in rain rate (Bretherton et al., 2010).
The NW model tends to exhibit better agreement with observations, showing lower mean and median rainfall rates.
The NW model has higher concentrations of activated particles and smaller effective droplet radius, which decrease
autoconversion and suppress precipitation (Albrecht, 1989).
The equilibrium reached in the base simulation is off since as
more precipitation is produced, more particles are scavenged
(and not recovered after evaporation), further reducing the
number of particles and leading to even more precipitation
in a reinforcing feedback. Besides showing higher precipitation rates, the base run also shows higher precipitation occurrence along the flight track (sampling time on the top of each
plot), while the NW results tend to agree more closely with
observations. At lower altitudes (500 and 100 m), observed
precipitation occurrence decreases, which is also captured by
both models, with the NW model always showing lower occurrence. At 100 m, the base model shows better agreement
with remote observed rain rates and NW overestimates the
mean and medium. Since the NW rain is scarcer, in agreement with observations, heavy drizzle events tend to skew
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
3055
g
(a)
(a)
(b)
(b)
(c)
(c)
(d)
(d)
Fig. 5. Box and whisker plots for selected cloud properties and
aerosol number concentration as in Fig. 1. (a) Profile mean cloud
water content. (b) Profile mean number of droplets concentration.
(c) and (d): marine boundary layer (MBL) and free troposphere
(FT) mean profile aerosol number concentration. Number of profiles is indicated at the top of each longitude bin.
the distribution. However, modeled rain range given by the
outer deciles agrees with the observations.
While episodic comparisons with in-situ observations are
critical, it is also important to consider model performance
for regional climatology, as the model should represent
monthly mean values and their spatial features. Figure 7
shows COD and LWP for MODIS and both WRF-Chem simulations. Model COD was computed by first computing the
effective radius as in Martin et al. (1994) and then COD as
proposed by Slingo (1989) for the 0.64–0.69 µm band, as the
MODIS reference wavelength for this retrieval is 0.65 µm
(King et al., 2006). The base WRF-Chem model usually
underestimates COD, while the NW model is closer to the
observations. Several features are well represented: close
to shore hotspots of COD around 17◦ S and 26◦ S, a nearshore local COD minimum around 36S, and an increase in
COD around 20◦ S from 80◦ W to 75◦ W. In the remote zone
(83◦ W to 90◦ W), observed COD tends to fall between both
models but closer to NW, for the same reasons presented
before to explain episodic performance: the base model is
www.atmos-chem-phys.net/12/3045/2012/
Fig. 6. Box and whisker plots for radar derived and modeled rain
at different heights as in Fig. 1. (a) and (b) corresponds to rain
rates just below the cloud top and at the cloud base while (c) and
(d) corresponds to rain rates at fixed heights of 500 and 100 m. The
numbers above each zone represent sampling time in decimal hours.
unable to generate a thick enough cloud layer and drizzles
too much, while the NW clouds do not dissipate when moving westwards, thereby increasing cloud lifetime (Albrecht,
1989). LWP path shows a different behavior, as both models
underestimate MODIS LWP, probably due to a model bias
in the Lin microphysics parameterization, as the Morrison
scheme (Morrison and Pinto, 2005) generates higher LWP
(Q. Yang et al., 2011), as discussed further in the text. However, NW model results consistently show higher values and
a better agreement with observations.
3.4
Aerosol feedbacks and relation to sources
The Ron Brown provided a unique platform for continuous
point measurements. Not only does the vessel have a much
longer residence time in each regional model grid-cell than
research aircraft, it also records the complete diurnal cycle
at each location. One of the points where Ron Brown sampling efforts were focused was at (20◦ S, 75◦ W), where it
spent approximately 8 days, 4 days on each leg of the cruise.
This area was found to be affected by coastal sources (Allen
Atmos. Chem. Phys., 12, 3045–3064, 2012
3056
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Fig. 7. MODIS-Aqua products and model monthly averages for the VOCALS period (15 October to 16 November). First and second row
show cloud optical depth (COD) and liquid water path (LWP), respectively while first, second and third columns show MODIS, the base
model and the no wet deposition (NW) model, respectively.
et al., 2011; Bretherton et al., 2010; Hawkins et al., 2010). To
evaluate model performance and aerosol interactions, Fig. 8a
compares total observed sulfate to the base and NW models.
Observed values are closer to the NW model, but both models resolve most of the periods where SO4 concentrations increase over 1 µg m−3 . As seen in Fig. 8b, the SO4 episodes
are well correlated with aerosol number concentrations over
13 nm in diameter, a relationship also represented by the
model. Observed aerosol number concentration is in the high
range of the models because the lowest bin modeled is 40 nm,
and does not include the 13 to 40 nm window. These sulfate
episodes do not follow any diurnal pattern, and are a constant
factor affecting aerosol concentrations in this zone. Model
results, including prior modeling by Spak et al. (2010) that
only included anthropogenic sulfur emissions, clearly indicate that these peaks can be attributed to continental sources,
usually coming from Central Chile. As an example, Fig. 8c
shows the evolution of second bin (78 to 156 nm in diameter) SO4 (main contributor to aerosol number concentration)
in a distinct pollution plume from the time emitted in Central Chile until it reaches the Ron Brown, 2 days after. When
fresh, the maximum value of the plume is found on model
level 17, around 650 m above sea level. At this height, it
is transported by southeasterly trades (Rahn and Garreaud,
2010) until it makes contact with the MBL, where it starts
entraining and SO2 to SO4 conversion is enhanced in clouds.
Once in the MBL, lower wind speeds result in the plume taking a longer time to reach the Ron Brown location. In the
MBL, the plume receives additional SO4 contributions from
DMS, as a near-shore DMS emissions hot spot is found off
central Chile (26◦ S-36◦ S) due to wind shear generated by
Atmos. Chem. Phys., 12, 3045–3064, 2012
the subtropical low-level jet (Garreaud and Muñoz, 2005;
Muñoz and Garreaud, 2005). By performing a simulation
without DMS initial conditions and emissions, we estimate
the DMS contribution to sulfate to be from 15 % to 25 %
in mass (which could be overestimated as shown in previous sections) by the time the plume reaches the Ron Brown
location for the case analyzed, showing that these episodes
are generated mainly by anthropogenic sources. The ability
of these plumes to reach this zone is thought to be determined by the position of the surface pressure maximum of
the Southeast Pacific Anticyclone (Spak et al., 2010).
Distant sources from Central Chile often have a visible
footprint in SO4 mass and aerosol number concentration over
the study domain, and these enhanced aerosols participate in
cloud feedbacks such as drizzle suppression. Figure 9 shows
curtain plots of radar reflectivity (proxy for precipitation) and
aerosol number concentration for C-130 RF05 flight, showing very marked and correlated longitudinal gradients both
on aerosol load and precipitation for NW and observed values. The base model (W) shows the same gradient (not
shown), with higher rain rates in the remote region. Bretherton et al. (2010) argued that lack of drizzle near the coast is
not just a microphysical response to high droplet concentrations; but other aspects such as lower LWP and thinner clouds
(related to shallower and coupled MBL) can be comparably important. However, synoptic conditions present during
RF05 flight were such that MBL height differences between
off shore and remote zone weren’t significant (cloud top and
base heights differences less than 250 m and 200 m respectively, Fig. 4) and the remote region was not decoupled (no
vertical gradients on aerosol concentrations inside the MBL,
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
3057
also verified by flight vertical profiles as in Fig. 3) but we
still see very high precipitation gradients. Thus, differences
in aerosol load might be playing a more important role than
previously thought. The enhanced aerosols also participate in
cloud feedbacks visible in satellite retrievals of cloud properties. Figure 10a shows MODIS-Aqua cloud effective radius for an overpass on a day with a thick cloud deck, where
aerosol feedbacks are more pronounced. Figure 10b, c show
model results for effective radius and second bin sulfate surface concentrations. Model cloud effective radius clearly decreases when the MBL is dominated by high accumulation
mode sulfate concentrations, following a similar shape to
the plume, which can also be observed in the MODIS overpass. The scene shows two distinct plumes coming from
Central Chile: an older one between 23◦ S and 20◦ S and
a fresher one in between 29◦ S and 25◦ S, both showing a decrease in cloud effective radius in both model and the observations. These findings highlights the need to consider
aerosol interactions and transport from far-away sources in
high-resolution studies and NWP applications over the region and similar persistent coastal stratocumulus in eastern
boundary tropical and subtropical areas.
3.5
Fig. 8. (a) Time series for observed and modeled SO4 . (b) Time
series for observed and modeled aerosol number concentration for
diameters over 13 nm and 40 nm respectively. Black thick lines in
(a) and (b) divides both periods that Ron Brown stayed 4 days on
75◦ W: 29 October–1 November and 11–15 November. (c) Composite of NW model second bin (78–156 nm aerosol diameter) SO4
concentration in µg m−3 . Each composite follows the same plume
since it is emitted on Central Chile until it reaches Ron Brown
(marked by a circle) two days after. The two most southern composites are extracted from level 17 (∼670 m over sea level) while the
rest are extracted from the first model level. Scale is logarithmic.
www.atmos-chem-phys.net/12/3045/2012/
Assessing differences due to model configuration
WRF-Chem is a community model with several choices
of parameterizations to represent various processes (Skamarock et al., 2008). These choices result in different
model configurations, which can produce different predictions. For example, there are two microphysics schemes
that can be used to study aerosol indirect effects in WRFChem v3.3: the Lin scheme (used in this study); and the
Morrison scheme (Q. Yang et al., 2011; Morrison and Pinto,
2005). Figure 11 shows results from a sensitivity column
study where both schemes were run until reaching stable
conditions as a function of number of droplets. Significant
differences of over an order of magnitude are found in rain
rates between both schemes. For the VOCALS-REx case of
study, the Lin scheme rain rates showed good performance
(Fig. 6), while the Morrison scheme showed under-prediction
(Q. Yang et al., 2011), in accordance with Fig. 11. The higher
rain rates in our study can also explain the larger underprediction of sulfate mass and aerosol number concentration
by the base model (Figs. 1 and 3) compared to the Q. Yang
et al. (2011) study. We hypothesize that the main cause of the
rain mismatch is the different autoconversion (cloud droplets
to rain conversion) parameterization used in both schemes.
The Morrison scheme uses Khairoutdinov and Kogan (2000)
parameterization, which uses regressed coefficients (multiplicative and power laws) from cloud drop size spectra predicted by LES simulations, which shows a linear behavior
in the log scale (Fig. 11a); while the Lin scheme uses Liu
et al. (2005) which introduces a threshold function which depends on droplet number concentration that is responsible
for the curve shape and rain suppression in Fig. 11a. On the
Atmos. Chem. Phys., 12, 3045–3064, 2012
3058
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Fig. 9. Curtain plots for radar reflectivity (Z, in dBZ) and accumulation mode aerosol number concentration (# cm−3 ) for C-130 flight RF05
on 25 October. (a) and (b) shows radar observed and NW model Z while (c) shows NW model as the curtain and one minute average PCASP
observations as colored circles. Observed Z and PCASP aerosol are 1 min averages. Model Z is computed according to Appendix A. Solid
lines represent flight track with the line becoming segmented on (c) every time there is a PCASP observation. For all panels, bottom scale is
time in hours and top scale is longitude in degrees.
Fig. 10. Horizontal plots of cloud effective radius (µm, a and b) and first level, second bin (78–156 nm aerosol diameter) SO4 concentration
(µg m−3 , c). (a) shows MODIS-AQUA cloud effective radius for 16 October 17:00 UTC overpass while (b) and (c) shows NW model results
for the same time. Model cloud effective radius is computed for the cloud top.
Atmos. Chem. Phys., 12, 3045–3064, 2012
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
(a)
g
(b)
Fig. 11. Results from column study for comparing Lin and Morrison microphysics schemes for a profile on (80◦ W, 20◦ S) at
00:00 UTC on 28 October 2008. (a) shows maximum rain rate per
profile while (b) shows liquid water path (LWP) per profile. Each
profile is run with a different droplet number concentration using
a 12 s time step for enough time to reach stable conditions. For (a),
missing points means rain rate equal to zero.
other hand, the Morrison scheme shows higher liquid water content (Fig. 11b), which is not completely explained by
the lower precipitation, as LWP differences are still present
when rain rates are similar for low number of droplets. This
is also coherent with the fact that Q. Yang et al. (2011) shows
better agreement to MODIS LWP than our study, where this
configuration under-predicts it.
Full double moment microphysics schemes (Lin scheme
is double moment for cloud water only) are necessary to improve process representation in models (e.g., Morrison et al.,
2009). As autoconversion seems to be generating low performance in rain rates, we propose to implement and test the
Liu et al. (2005) parameterization in the Morrison scheme.
The implementation has to come along with the inclusion
of aerosol re-suspension due to rain evaporation on the wet
deposition scheme to avoid the MBL aerosol biases seen in
Figs. 1 and 3.
4
Conclusions
There is an imperative need for reducing uncertainty and improving the atmospheric models used in studies of aerosolcloud interactions at scales needed for NWP, air quality predictions, and policy assessments. In this context, several
intensive measurement campaigns have been carried out to
improve our understanding of aerosol and cloud interactions
and they provide an extensive data base for use in testing
and improving models. In this work we test the regional
model WRF-Chem for the VOCALS-REx campaign which
focused on studying the persistent stratocumulus deck on the
South East Pacific, off the shore of Chile and Peru. Starting from the fact that the inclusion of aerosol cloud interwww.atmos-chem-phys.net/12/3045/2012/
3059
actions in the model are important to represent processes in
this region (Q. Yang et al., 2011), we perform model simulations designed to address the questions: what are the effects
on cloud dynamics and microphysics from changing the subcloud aerosol loads? And do these effects bring model results
closer to observations when aerosol loads are in better agreement to measurements? To address these questions results
from two model simulations, with (base) and without wet deposition (NW) were analyzed. Both runs represent an incomplete modeling picture, as the base run lacks aerosol resuspension (which is important in drizzling stratocumulus), and
excluding wet deposition means neglecting a known removal
process. These simulations produce significant differences
in aerosol amounts, particularly in the remote zone where
sulfate mass and accumulation mode aerosol number distributions do not overlap with each other and can be one order of magnitude different. Observed aerosol mass and number are usually closer to the NW results, because the model
wet deposition process irreversibly removes aerosol even for
evaporating rain. Little surface rain was observed during the
campaign, so evaporation of drizzle drops is a likely source
of sub-cloud aerosols. The increase in aerosol number in
NW generate a significant difference between the models in
terms of marine boundary layer (MBL) dynamics and cloud
microphysics, in accordance to warm clouds aerosol indirect
effects. These include an increased number of cloud droplets
(Twomey, 1974) showing no overlap of the inner quartiles
from the two models in the remote zone; increased MBL
and cloud heights (Pincus and Baker, 1994) reaching up to
200 m differences; drizzle suppression on average concentrations and on number of detections; increased liquid water
content and increased cloud lifetime (Albrecht, 1989); which
helps answer the first question. MBL dynamics and cloud
microphysics observed values are usually closer to NW or at
least fall in between both models showing that better aerosol
statistical performance lead to changes in the right direction,
which helps answer the second question. This study demonstrates the capabilities of the WRF-Chem model to simulate
aerosol/cloud interactions, particularly regarding the activation routine, which simulates number of droplet concentrations more accurately when sub-cloud aerosol loads more
closely match observations. However, the model needs further improvements to address issues such as aerosol resuspension in rain wet removal, overestimation in oceanic DMS
and sea-salt emissions, increased cloud driven SO2 to sulfate
conversion and move from bulk to sectional/modal aqueous
chemistry. Also, an assessment of model differences when
using distinct WRF-Chem configurations shows these seem
to be related to the microphysics schemes, specifically to
different autoconversion parameterizations which can generate over an order of magnitude disagreement on rain rates
predictions for the same conditions.
Besides performing campaign averaged comparisons,
we quantified local model performance for stratocumulus
properties and their hourly evolution against ship-based
Atmos. Chem. Phys., 12, 3045–3064, 2012
3060
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
measurements (NOAA Ron Brown), three aircraft observations (NSF C-130, DoE G-1, FAAM BAe-146) and satellite retrievals (MODIS) using them to explain how aerosols
and model processes affect system response. For instance,
hourly evolution of cloud heights was evaluated showing a
good model performance for the diurnal cycle and different
synoptic conditions with the exception of periods where the
model is not able to recover from the underestimated MBL
height found on the boundary conditions. Also, an episodic
study was performed showing that anthropogenic sources
from Central Chile substantially changed aerosol mass and
number, rain and cloud optical properties over the ocean both
in modeled and observed values, showing that indirect effects
might be playing a more important role in modulating cloud
properties and dynamics than stated in previous studies.
In our analysis we have attempted to perform a complete
multi-platform evaluation for a regional simulation of clouds
and aerosols, where we included VOCALS-REx observations which were not compared to models previously, such
as decoupling state, trace gas concentrations (carbon monoxide, ozone), cloud aerosol composition, cloud water ionic
balance and radar reflectivities. These are all crucial to fully
quantifying regional model performance in this tightly coupled system. In order to provide quantification to this evaluation, we introduced a new metric for assessing model performance that uses box and whisker plots. This metric is independent of the variable being analyzed, thus allows doing
performance cross-comparison in between different models
and variables.
Together with improving model issues already mentioned,
future work should be focused on continuing to validate models with aerosol and cloud interactions from measurement
campaigns in other locations, as conditions for each region
vary extensively. Also, several observation platforms such
as close to shore flights (NERC Dornier 228 and CIRPAS
Twin Otter) and inland measurements (Iquique, Paposo and
Paranal sites) were not considered as the modeling was too
coarse for their use (12 km2 grid cells). Thus, finer resolution studies for the same area are needed to exploit these
data (4–1 km2 grid cells). These studies help to better quantify the uncertainties in models, that need to be considered
when these models are used as tools for policy makers and
for weather and air quality forecasts.
Appendix A
Model radar reflectivity
The Lin microphysics scheme implemented in WRF-Chem
uses an exponential distribution for rain droplets (Chen and
Sun, 2002):
N(D) = N0 exp(−3D)
Atmos. Chem. Phys., 12, 3045–3064, 2012
3=
πρw N0
ρqr
1/4
Where N0 is the intercept parameter, 3 the slope parameter,
ρw and ρ water and air densities respectively, and qr the rain
water content. Radar reflectivity (Z) is computed as the sixth
moment of the rain drop distribution as:
Z ∞
Z=
D 6 N (D)dD
0
which is only valid for Rayleigh scattering regime. As
the radar used in this study is W-band (94 GHz frequency,
∼3 mm wavelength) then Rayleigh scattering might not be
valid for droplets over 0.5 mm (O’Connor et al., 2004). Instead, we use Mie calculations to obtain Z (Arai et al., 2005):
Z ∞
λ4
σMie N (D)dD
ZMie = 5 2
(A2)
π |K| 0
Where λ is the radar wavelength, K the absorption coefficient of water and σMie the backscattering cross-section,
which is a function of droplet diameter and radar wavelength.
Then, as N0 is fixed for the Lin scheme (8e+6m4 ), ZMie can
be computed as a function of rain water content by numerically integrating this equation. For each diameter, σMie is
computed using Mätzler (2002) code which is based on the
Appendix of Bohren and Huffman (1983).
Acknowledgements. We thank all people and organizations that
participated in the VOCALS-REx campaign, generating and
allowing us to use the complete and comprehensive data set.
Special thanks to Timothy Bates, Paquita Zuidema and Ludovic
Bariteau for helping interpret and compare observational data, to
William Gustafson and Jerome Fast for comments on modeling
and to two anonymous reviewers for their constructive comments.
Ron Brown, C-130 and G-1 measurements were obtained from the
VOCALS data archive NCAR/EOL, which is sponsored by the
National Science Foundation (NSF). MODIS data was obtained
from the NASA Langley Research Center Atmospheric Science
Data Center. Special thanks to the staff of the NCAR Research
Aviation Facility for supporting the deployment of the C130 and
the staff of the UWYO King Air national facility for enabling
the deployment of the WCR and WCL onboard the C130 during
VOCALS-REx. We also thank the UK Natural Environment Research Council (NERC) for funding the VOCALS UK contingent
to the project (grant ref: NE/F019874/1) and the NERC Facility
for Airborne and Atmospheric Measurment (FAAM) and Direct
Flight and Avalon for operational support of the BAe-146 aircraft.
This work was carried out with the aid of NSF grants 0748012 and
0745986, grant number UL1RR024979 from the National Center
for Research Resources (NCRR), a part of the National Institutes
of Health (NIH), FONDECYT Iniciación grant 11090084, and
Fulbright-CONICYT scholarship number 15093810. Its contents
are solely the responsibility of the authors and do not necessarily
represent the official views of the founding institutions.
Edited by: R. Wood
(A1)
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
References
Abdul-Razzak, H. and Ghan, S. J.: A parameterization of aerosol
activation. 3. Sectional representation, J. Geophys. Res., 107,
4026, doi:10.1029/2001JD000483, 2002.
Abel, S. J., Walters, D. N., and Allen, G.: Evaluation of stratocumulus cloud prediction in the Met Office forecast model
during VOCALS-REx, Atmos. Chem. Phys., 10, 10541–10559,
doi:10.5194/acp-10-10541-2010, 2010.
Albrecht, B. A.: Aerosols, cloud microphysics, and fractional
cloudiness, Science, 245, 1227–1230, 1989.
Allen, G., Coe, H., Clarke, A., Bretherton, C., Wood, R., Abel, S. J.,
Barrett, P., Brown, P., George, R., Freitag, S., McNaughton, C.,
Howell, S., Shank, L., Kapustin, V., Brekhovskikh, V., Kleinman, L., Lee, Y.-N., Springston, S., Toniazzo, T., Krejci, R.,
Fochesatto, J., Shaw, G., Krecl, P., Brooks, B., McMeeking, G.,
Bower, K. N., Williams, P. I., Crosier, J., Crawford, I., Connolly, P., Allan, J. D., Covert, D., Bandy, A. R., Russell, L. M.,
Trembath, J., Bart, M., McQuaid, J. B., Wang, J., and Chand, D.:
South East Pacific atmospheric composition and variability sampled along 20◦ S during VOCALS-REx, Atmos. Chem. Phys.,
11, 5237–5262, doi:10.5194/acp-11-5237-2011, 2011.
Andrejczuk, M., Grabowski, W. W., Gadian, A., and Burton, R.: Limited-area modelling of stratocumulus over SouthEastern Pacific, Atmos. Chem. Phys. Discuss., 11, 25517–25556,
doi:10.5194/acpd-11-25517-2011, 2011.
Andronache, C., Donner, L. J., Seman, C. J., Ramaswamy, V., and
Hemler, R. S.: Atmospheric sulfur and deep convective clouds
in tropical Pacific: a model study, J. Geophys. Res., 104, 4005–
4024, 1999.
Arai, K., Liang, X. M., and Liu, Q.: Method for estimation of rain
rate with Rayleigh and Mie scattering assumptions on the ZR relationship for different rainfall types, Adv. Space Res., 36,
813–817, 2005.
Bator, A. and Collett Jr., J. L.: Cloud chemistry varies
with drop size, J. Geophys. Res., 102, 28071–28078,
doi:10.1029/97JD02306, 1997
Benedict, K. B., Lee, T., and Collett Jr., J. L.: Cloud water
composition over the Southeastern Pacific Ocean during the
VOCALS Regional Experiment, Atmos. Environ.,46, 104–114,
doi:10.1016/j.atmosenv.2011.10.029, 2012.
Berner, A. H., Bretherton, C. S., and Wood, R.: Large-eddy simulation of mesoscale dynamics and entrainment around a pocket
of open cells observed in VOCALS-REx RF06, Atmos. Chem.
Phys., 11, 10525–10540, doi:10.5194/acp-11-10525-2011, 2011.
Bohren, C. F. and Huffman, D. R.: Absorption of Light by Small
Particles, Wiley, New York, 1983.
Bond, T. C., Streets, D. G., Yarber, K. F., Nelson, S. M., Woo, J.H., and Klimont, Z.: A technology-based global inventory of
black and organic carbon emissions from combustion, J. Geophys. Res., 109, D14203, doi:10.1029/2003JD003697, 2004.
Bretherton, C. S., Wood, R., George, R. C., Leon, D., Allen, G.,
and Zheng, X.: Southeast Pacific stratocumulus clouds, precipitation and boundary layer structure sampled along 20◦ S during VOCALS-REx, Atmos. Chem. Phys., 10, 10639–10654,
doi:10.5194/acp-10-10639-2010, 2010.
Cess, R. D., Potter, G. L., Blanchet, J. P., Boer, G. J., Ghan, S. J.,
Kiehl, J. T., Le Treut, H., Li, Z. X., Liang, X. Z., Mitchell,
J. F. B., and others: Interpretation of cloud-climate feedback as
produced by 14 atmospheric general circulation models, Science,
www.atmos-chem-phys.net/12/3045/2012/
3061
245, 513–516, 1989.
Chand, D., Hegg, D. A., Wood, R., Shaw, G. E., Wallace, D.,
and Covert, D. S.: Source attribution of climatically important
aerosol properties measured at Paposo (Chile) during VOCALS,
Atmos. Chem. Phys., 10, 10789–10801, doi:10.5194/acp-1010789-2010, 2010.
Chapman, E. G., Gustafson Jr., W. I., Easter, R. C., Barnard, J. C.,
Ghan, S. J., Pekour, M. S., and Fast, J. D.: Coupling aerosolcloud-radiative processes in the WRF-Chem model: Investigating the radiative impact of elevated point sources, Atmos. Chem.
Phys., 9, 945–964, doi:10.5194/acp-9-945-2009, 2009.
Chen, S.-H. and Sun, W.- Y.: A one-dimensional time dependent
cloud model, J. Meteorol. Soc. Japan, 80, 99–118, 2002.
Chen, Y.-C., Xue, L., Lebo, Z. J., Wang, H., Rasmussen, R. M., and
Seinfeld, J. H.: A comprehensive numerical study of aerosolcloud-precipitation interactions in marine stratocumulus, Atmos.
Chem. Phys., 11, 9749–9769, doi:10.5194/acp-11-9749-2011,
2011.
Chou, M. D., Suarez, M. J., Ho, C. H., Yan, M. M. H., and
Lee, K. T.: Parameterizations for cloud overlapping and shortwave single-scattering properties for use in general circulation
and cloud ensemble models, J. Climate, 11, 202–214, 1998.
Comstock, K. K., Wood, R., Yuter, S. E., and Bretherton, C. S.: Reflectivity and rain rate in and below drizzling stratocumulus, Q. J.
Roy. Meteorol. Soc., 130, 2891–2918, 2004.
Corvalán, R. M., Osses, M., Urrutia, C. M., Gonzalez, P. A.: Estimating traffic emissions using demographic and socio-economic
variables in 18 Chilean urban areas, Popul. Environ., 27, 63–87,
2005.
Davies, D. K., Ilavajhala, S., Wong, M. M., and Justice, C. O.: Fire
information for resource management system: archiving and distributing MODIS active fire data, IEEE T. Geosci. Remote Sens.,
47, 72–79, doi:10.1109/TGRS.2008.2002076, 2009.
de Szoeke, S. P., Fairall, C. W., Wolfe, D. E., Bariteau, L., and
Zuidema, P.: Surface flux observations on the southeastern tropical pacific ocean and attribution of SST errors in coupled oceanatmosphere models, J. Climate, 23, 4152–4174, 2010.
DICTUC: Actualización del Invexntario de Emisiones de Contaminantes Atmosféricos en la Región Metropolitana 2005, Prepared
for Chilean Ministry of Environment, Santiago, Chile, 2007.
Easter, R. C., Ghan, S. J., Zhang, Y., Saylor, R. D., Chapman, E. G.,
Laulainen, N. S., Abdul-Razzak, H., Leung, L. R., Bian, X.,
and Zaveri, R. A.: MIRAGE: Model description and evaluation of aerosols and trace gases, J. Geophys. Res., 109, D20210,
doi:10.1029/2004JD004571, 2004.
Emmons, L. K., Walters, S., Hess, P. G., Lamarque, J.-F., Pfister, G. G., Fillmore, D., Granier, C., Guenther, A., Kinnison, D.,
Laepple, T., Orlando, J., Tie, X., Tyndall, G., Wiedinmyer, C.,
Baughcum, S. L., and Kloster, S.: Description and evaluation of
the Model for Ozone and Related chemical Tracers, version 4
(MOZART-4), Geosci. Model Dev., 3, 43–67, doi:10.5194/gmd3-43-2010, 2010.
Errico, R. M., Bauer, P., and Mahfouf, J. F.: Issues regarding the
assimilation of cloud and precipitation data, J. Atmos. Sci., 64,
3785–3798, 2007.
Fahey, K. M. and Pandis, S. N.: Optimizing model performance:
variable size resolution in cloud chemistry modeling, Atmos. Environ., 35, 4471–4478, 2001.
Fast, J. D., Gustafson Jr., W. I., Easter, R. C., Zaveri, R. A.,
Atmos. Chem. Phys., 12, 3045–3064, 2012
3062
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Barnard, J. C., Chapman, E. G., Grell, G. A., and Peckham, S. E.: Evolution of ozone, particulates, and aerosol direct
radiative forcing in the vicinity of Houston using a fully coupled
meteorology-chemistry-aerosol model, J. Geophys. Res., 111,
D21305, doi:10.1029/2005JD006721, 2006.
Feingold, G., Kreidenweis, S. M., Stevens, B., and Cotton, W. R.:
Numerical simulations of stratocumulus processing of cloud
condensation nuclei through collision-coalescence, J. Geophys.
Res., 101, 21391–21402, doi:10.1029/96JD01552, 1996.
Feingold, G., Walko, R. L., Stevens, B., and Cotton, W. R.: Simulations of marine stratocumulus using a new microphysical parameterization scheme, Atmos. Res., 47–48, 505–528, 1998.
Freitas, S. R., Longo, K. M., and Andreae, M. O.: Impact of including the plume rise of vegetation fires in numerical simulations
of associated atmospheric pollutants, Geophys. Res. Lett., 33,
L17808, doi:10.1029/2006GL026608, 2006.
Freitas, S. R., Longo, K. M., Chatfield, R., Latham, D.,
Silva Dias, M. A. F., Andreae, M. O., Prins, E., Santos, J. C.,
Gielow, R., and Carvalho Jr., J. A.: Including the sub-grid
scale plume rise of vegetation fires in low resolution atmospheric transport models, Atmos. Chem. Phys., 7, 3385–3398,
doi:10.5194/acp-7-3385-2007, 2007.
Garreaud, R. D. and Muñoz, R. C.: The low-level jet off the west
coast of subtropical South America: structure and variability,
Mon. Weather Rev., 133, 2246–2261, 2005.
Ghan, S. J. and Easter, R. C.: Impact of cloud-borne aerosol representation on aerosol direct and indirect effects, Atmos. Chem.
Phys., 6, 4163–4174, doi:10.5194/acp-6-4163-2006, 2006.
Gong, S., Barrie, L. A., and Blanchet, J.-P.: Modeling sea salt
aerosols in the atmosphere. 1: Model development, J. Geophys.
Res., 102, 3805–3818, 1997.
Graßl, H.: Possible changes of planetary albedo due to aerosol
particles, in: Man’s Impact on Climate, edited by: Bach, W.,
Pankrath, J., and Kellogg, W., Elsevier, New York, 1979.
Grell, G. A., Peckham, S. E., Schmitz, R., McKeen, S. A., Frost, G.,
Skamarock, W. C., and Eder, B.: Fully coupled “online” chemistry within the WRF model, Atmos. Environ. 39, 6957–6975,
2005.
Guenther, A., Karl, T., Harley, P., Wiedinmyer, C., Palmer, P. I.,
and Geron, C.: Estimates of global terrestrial isoprene emissions
using MEGAN (Model of Emissions of Gases and Aerosols from
Nature), Atmos. Chem. Phys., 6, 3181–3210, doi:10.5194/acp-63181-2006, 2006.
Gustafson Jr., W. I., Chapman, E. G., Ghan, S. J., Easter, R. C.,
and Fast, J. D.: Impact on modeled cloud characteristics
due to simplified treatment of uniform cloud condensation nuclei during NEAQS 2004, Geophys. Res. Lett., 34, L19809,
doi:10.1029/2007GL030021, 2007.
Hawkins, L. N., Russell, L. M., Covert, D. S., Quinn, P. K., and
Bates, T. S.: Carboxylic acids, sulfates, and organosulfates in
processed continental organic aerosol over the Southeast Pacific Ocean during VOCALS-REx 2008, J. Geophys. Res., 115,
D13201, doi:10.1029/2009JD013276, 2010.
Hind, A. J., Rauschenberg, C. D., Johnson, J. E., Yang, M., and
Matrai, P. A.: The use of algorithms to predict surface seawater
dimethyl sulphide concentrations in the SE Pacific, a region of
steep gradients in primary productivity, biomass and mixed layer
depth, Biogeosciences, 8, 1–16, doi:10.5194/bg-8-1-2011, 2011.
Hoffmann, M. R. and Calvert, J. G.: Chemical transformation
Atmos. Chem. Phys., 12, 3045–3064, 2012
modules for Eulerian acid deposition models, Vol. 2, in: The
Aqueous-Phase Chemistry, EPA/600/3-85/017, US Environmental Protection Agency, Research Triangle Park, NC, 1985.
Jones, C. R., Bretherton, C. S., and Leon, D.: Coupled vs. decoupled boundary layers in VOCALS-REx, Atmos. Chem. Phys.,
11, 7143–7153, doi:10.5194/acp-11-7143-2011, 2011.
Jorquera, H. and Castro, J.: Analysis of urban pollution episodes by
inverse modeling, Atmos. Environ. 44, 42–54, 2010.
Kanakidou, M., Seinfeld, J. H., Pandis, S. N., Barnes, I., Dentener, F. J., Facchini, M. C., Van Dingenen, R., Ervens, B.,
Nenes, A., Nielsen, C. J., Swietlicki, E., Putaud, J. P., Balkanski, Y., Fuzzi, S., Horth, J., Moortgat, G. K., Winterhalter, R.,
Myhre, C. E. L., Tsigaridis, K., Vignati, E., Stephanou, E. G.,
and Wilson, J.: Organic aerosol and global climate modelling: a
review, Atmos. Chem. Phys., 5, 1053–1123, doi:10.5194/acp-51053-2005, 2005.
KAS: “Análisis General del Impacto Social y Económico de la
Norma de Emisión Para Termoeléctricas.” Chilean Ministry
of the Environment, available at: http://www.sinia.cl/1292/
articles-44963 informe final term.pdf, last access: November
2011, 2009.
Kazil, J., Wang, H., Feingold, G., Clarke, A. D., Snider, J. R., and
Bandy, A. R.: Modeling chemical and aerosol processes in the
transition from closed to open cells during VOCALS-REx, Atmos. Chem. Phys., 11, 7491–7514, doi:10.5194/acp-11-74912011, 2011.
Khairoutdinov, M. and Kogan, Y.: A new cloud physics parameterization in a large-eddy simulation model of marine stratocumulus,
Mon. Weather Rev. 128, 229–243, 2000.
King, M. D., Platnick, S., Hubanks, P. A., Arnold, G. T.,
Moody, E. G., Wind, G., and Wind, B.: Collection 005 change
summary for the MODIS cloud optical property (06 OD) algorithm, available at: http://modis-atmos.gsfc.nasa.gov/C005
Changes/C005 CloudOpticalProperties ver311.pdf, last accesse:
November 2011, 2006.
Kleinman, L. I., Daum, P. H., Lee, Y.-N., Lewis, E. R., Sedlacek
III, A. J., Senum, G. I., Springston, S. R., Wang, J., Hubbe, J.,
Jayne, J., Min, Q., Yum, S. S., and Allen, G.: Aerosol concentration and size distribution measured below, in, and above cloud
from the DOE G-1 during VOCALS-REx, Atmos. Chem. Phys.,
12, 207–223, doi:10.5194/acp-12-207-2012, 2012.
Koren, I., Martins, J. V., Remer, L. A., and Afargan, H.: Smoke invigoration versus inhibition of clouds over the Amazon, Science,
321, 946–949, 2008.
Liepert, B. G.: Observed reductions of surface solar radiation at
sites in the United States and worldwide from 1961 to 1990, Geophys. Res. Lett., 29, 1421, doi:10.1029/2002GL014910, 2002.
Liss, P. S. and Merlivat, L.: Air-sea gas exchange rates: Introduction
and synthesis, in: The role of air-sea exchange in geochemical
cycling, edited by: Buat-Menard, P., 113–127, 1986.
Liu, Y., Daum, P. H., and McGraw, R. L.: Size truncation effect, threshold behavior, and a new type of autoconversion parameterization, Geophys. Res. Lett., 32, L11811,
doi:10.1029/2005GL022636, 2005.
Lohmann, U.: A glaciation indirect aerosol effect caused
by soot aerosols, Geophys. Res. Lett., 29, 1052,
doi:10.1029/2001GL014 357, 2002.
www.atmos-chem-phys.net/12/3045/2012/
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Lohmann, U. and Feichter, J.: Global indirect aerosol effects: a review, Atmos. Chem. Phys., 5, 715–737, doi:10.5194/acp-5-7152005, 2005.
Lohmann, U. and Lesins, G.: Stronger constraints on the anthropogenic indirect aerosol effect, Science, 298, 1012–1016, 2002.
Mätzler, C.: MATLAB Functions for Mie Scattering and Absorption, Institut für Angewandte Physik, Research Report No. 2002–
08, Bern, Switzerland, 2002.
Martin, G. M., Johnson, W., and Spice, A.: The measurement and
parameterization of effective radius of droplets in warm stratocumulus, J. Atmos. Sci., 51, 1823–1842, 1994.
Martin, L. and Damschen, D. E.: Aqueous oxidation of sulfur dioxide by hydrogen peroxide at low pH, Atmos. Environ., 15, 1615–
1621, doi:10.1016/0004-6981(81)90146-3, 1981.
McArdle, J. V. and Hoffmann, M. R.: Kinetics and mechanism of
the oxidation of aquated sulfur dioxide by hydrogen peroxide at
low pH, J. Phys. Chem., 87, 5425–5429, 1983.
McFarquhar, G. M., Um, J., Freer, M., Baumgardner, D.,
Kok, G. L., and Mace, G.: The importance of small ice crystals to cirrus properties: observations from the Tropical Warm
Pool International cloud Experiment (TWP-ICE), Geophys. Res.
Lett., 57, L13803, doi:10.1029/2007GL029865, 2007.
Mena-Carrasco, M., Ruiz, P., Stambuk, A., Stambuk, L., Granifo, R., and Jerardino, M.: Antecedentes para la Revisión de las
Normas Primarias de Calidad de Aire para Dióxido de Azufre,
Monóxido de Carbono, Ozono, Dióxido de Nitrógeno. Chilean
Ministry of the Environment, Santiago, Chile, 2010.
Mena-Carrasco, M., Oliva, E., Saide, P., Spak, S., de la Maza, C.,
Osses, M., Tolvett, S., Molina, L. T.: Estimating social and economic benefits from natural gas use in transport and heating in
Santiago, Chile, submitted, 2012.
Michel, Y. and Auligné, T.: Inhomogeneous background error modeling and estimation over Antarctica, Mon. Weather Rev., 138,
2229–2252, 2010.
Mitra, S. K., Brinkmann, J., and Pruppacher, H. R.: A wind tunnel study on the drop-to-particle conversion, J. Aerosol Sci., 23,
245–256, doi:10.1016/0021–8502(92)90326-Q, 1992.
Mlawer, E. J., Taubman, S. J., Brown, P. D., Iacono, M. J., and
Clough, S. A.: Radiative transfer for inhomogeneous atmosphere: RRTM, a validated correlated-k model for the longwave,
J. Geophys. Res., 102, 16663–16682, 1997.
Morrison, H. and Pinto, J. O.: Mesoscale modeling of springtime
Arctic mixed-phase stratiform clouds using a new two-moment
bulk microphysics scheme, J. Atmos. Sci., 62, 3683–3704, 2005.
Morrison, H., Thompson, G., and Tatarskii, V.: Impact of cloud
microphysics on the development of trailing stratiform precipitation in a simulated squall line: comparison of one- and
two-moment schemes, Mon. Weather Rev., 137, 991–1007,
doi:10.1175/2008mwr2556.1, 2009.
Muñoz, R. C. and Garreaud, R. D.: Dynamics of the low-level jet
off the west coast of subtropical South America, Mon. Weather
Rev., 133, 3661–3677, 2005.
Nakanishi, M. and Niino, H.: An improved Mellor-Yamada level3 model with condensation physics: its design and verification,
Bound.-Lay. Meteorol. 112, 1–31, 2004.
Ntelekos, A. A., Smith, J. A., Donner, L. J., Fast, J. D., Gustafson
Jr., W. I., Chapman, E. G., and Krajewski, W. F.: The effects of
aerosols on intense convective precipitation in the Northeastern
US, Q. J. Roy. Meteor. Soc., 135, 1367–1391, 2009.
www.atmos-chem-phys.net/12/3045/2012/
3063
O’Connor, E. J., Illingworth, A. J., and Hogan, R. J.: A technique
for autocalibration of cloud lidar, J. Atmos. Oceanic Technol.,
21, 777–786, 2004.
Olivier, J. G. J., Bouwman, A. F., Van der Maas, C. W. M., and
Berdowski, J. J. M.: Emission Database for Global Atmospheric
Research (EDGAR), Environ. Monit. Assess., 31, 93–106, 1994.
Otkin, J. A. and Greenwald, T. J.: Comparison of WRF modelsimulated and MODIS-derived cloud data, Mon. Weather Rev.,
136, 1957–1970, 2008.
Otkin, J. A., Greenwald, T. J., Sieglaff, J., and Huang, H.-L.: Validation of a large-scale simulated brightness temperature dataset
using SEVIRI satellite observations, J. Appl. Meteor. Climatol.,
48, 1613–1626, doi:10.1175/2009JAMC2142.1, 2009.
Pincus, R. and Baker, M.: Effect of precipitation on the albedo susceptibility of clouds in the marine boundary layer, Nature, 372,
250–252, 1994.
Rahn, D. A. and Garreaud, R.: Marine boundary layer over the subtropical southeast Pacific during VOCALS-REx – Part 1: Mean
structure and diurnal cycle, Atmos. Chem. Phys., 10, 4491–4506,
doi:10.5194/acp-10-4491-2010, 2010.
Ramanathan, V., Cess, R. D., Harrison, E. F., Minnis, P., Barkstrom, B. R., Ahmad, E., and Hartmann, D.: Cloud-radiative
forcing and climate: results from the earth radiation budget experiment, Science, 243, 57–63, 1989.
Saide, P., Osses, A., Gallardo, L., and Osses, M.: Adjoint inverse
modeling of a CO emission inventory at the city scale: Santiago
de Chile’s case, Atmos. Chem. Phys. Discuss., 9, 6325–6361,
doi:10.5194/acpd-9-6325-2009, 2009.
Saide, P. E., Carmichael, G. R., Spak, S. N., Gallardo, L.,
Osses, A. E., Mena-Carrasco, M. A., and Pagowski, M.: Forecasting urban PM10 and PM2.5 pollution episodes in very
stable nocturnal conditions and complex terrain using WRFChem CO tracer model, Atmos. Environ., 45, 2769–2780,
doi:10.1016/j.atmosenv.2011.02.001, 2011.
Seinfeld, J. and Pandis, S.: Atmospheric Chemistry and Physics,
2nd edn., Wiley, Hoboken, New Jersey, USA, 478–483, 2006.
Shah, S., Rao, B. M., Kumar, P., and Pal, P. K.: Verification of
cloud cover forecast with INSAT observation over Western India, J. Earth Syst. Sci., 119, 775–781, 2010.
Shank, L. M., Howell, S., Clarke, A. D., Freitag, S.,
Brekhovskikh, V., Kapustin, V., McNaughton, C., Campos, T.,
and Wood, R.: Organic matter and non-refractory aerosol over
the remote Southeast Pacific: oceanic and combustion sources,
Atmos. Chem. Phys., 12, 557–576, doi:10.5194/acp-12-5572012, 2012.
Skamarock, W. C., Klemp, J. B., Dudhia, J., Gill, D. O.,
Barker, D. M., Duda, M. G., Huang, X.- Y., Wang, W., and Powers, J. G.: A Description of the Advanced Research WRF, Version 3. Technical Report NCAR/TN475+ST R. National Center
for Atmospheric Research Technical Note, Boulder, Colorado,
2008.
Slingo, A.: A GCM parameterization for the shortwave radiative
properties of water clouds, J. Atmos. Sci., 46, 1419–1427, 1989.
Solomon, S., Qin, D., Manning, M., Alley, R. B., Berntsen, T.,
Bindoff, N. L., Chen, Z., Chidthaisong, A., Gregory, J. M.,
Hegerl, G. C., Heimann, M., Hewitson, B., Hoskins, B. J.,
Joos, F., Jouzel, J., Kattsov, V., Lohmann, U., Matsuno, T.,
Molina, M., Nicholls, N., Overpeck, J., Raga, G., Ramaswamy, V., Ren, J., Rusticucci, M., Somerville, R.,
Atmos. Chem. Phys., 12, 3045–3064, 2012
3064
P. E. Saide et al.: Evaluating WRF-Chem aerosol indirect effects
Stocker, T. F., Whetton, P., Wood, R. A., and Wratt, D.: Technical Summary, in: Climate Change 2007: The Physical Science
Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change,
edited by: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K. B., Tignor, M., and Miller, H. L., Cambridge
University Press, Cambridge, New York, 2007.
Spak, S. N., Mena-Carrasco, M. A., and Carmichael, G. R.: Atmospheric transport of anthropogenic oxidized sulfur over the
Southeast Pacific during VOCALS REx, CLIVAR Exchanges,
53, 20–21, 2010.
Sun, R., Moorthi, S., Xiao, H., and Mechoso, C. R.: Simulation of
low clouds in the Southeast Pacific by the NCEP GFS: sensitivity to vertical mixing, Atmos. Chem. Phys., 10, 12261–12272,
doi:10.5194/acp-10-12261-2010, 2010.
Twomey, S.: Pollution and the planetary albedo, Atmos. Environ.,
8, 1251–1256, 1974.
Twomey, S.: Aerosols, clouds and radiation, Atmos. Environ. A,
25, 2435–2442, 1991.
Vellore, R., Koracin, D., and Wetzel, M.: A method of improving cloud predictions for real-time forecasting and visualization. Advances in Visual Computing, edited by: Bebis, G.,
Boyle, R., Parvin, B., Koracin, D., Remagnino, P., Nefian, A.,
Meenakshisundaram, G., Pascucci, V., Zara, J., Molineros, J.,
Theisel, H. and Malzbender, T., Springer, Berlin, Germany, 544–
553, 2006.
Vellore, R., Koracin, D., Wetzel, M., Chai, S., and Wang, Q.: Challenges in Mesoscale prediction of a nocturnal stratocumulustopped marine boundary layer and implications for operational
forecasting, Weather Forecast., 22, 1101–1122, 2007.
Wang, H. and Feingold, G.: Modeling mesoscale cellular structures
and drizzle in marine stratocumulus. Part I: Impact of drizzle on
the formation and evolution of open cells, J. Atmos. Sci., 66,
3257–3275, 2009a.
Wang, H. and Feingold, G.: Modeling mesoscale cellular structures and drizzle in marine stratocumulus. Part II: The microphysics and dynamics of the boundary region between
open and closed cells, J. Atmos. Sci., 66, 3257–3275,
doi:10.1175/2009JAS3120.1, 2009b.
Wang, H., Feingold, G., Wood, R., and Kazil, J.: Modelling microphysical and meteorological controls on precipitation and cloud
cellular structures in Southeast Pacific stratocumulus, Atmos.
Chem. Phys., 10, 6347–6362, doi:10.5194/acp-10-6347-2010,
2010.
Wang, S., O’Neill, L. W., Jiang, Q., de Szoeke, S. P., Hong, X.,
Jin, H., Thompson, W. T., and Zheng, X.: A regional real-time
forecast of marine boundary layers during VOCALS-REx, Atmos. Chem. Phys., 11, 421–437, doi:10.5194/acp-11-421-2011,
2011.
Wexler, A. S., Lurmann, F. W., and Seinfeld, J. H.: Modelling urban
and regional aerosols – I. Model development, Atmos. Environ.,
28, 531–546, 1994.
Atmos. Chem. Phys., 12, 3045–3064, 2012
Wood, R., Mechoso, C. R., Bretherton, C. S., Weller, R. A.,
Huebert, B., Straneo, F., Albrecht, B. A., Coe, H., Allen, G.,
Vaughan, G., Daum, P., Fairall, C., Chand, D., Gallardo Klenner, L., Garreaud, R., Grados, C., Covert, D. S., Bates, T. S.,
Krejci, R., Russell, L. M., de Szoeke, S., Brewer, A., Yuter, S. E.,
Springston, S. R., Chaigneau, A., Toniazzo, T., Minnis, P.,
Palikonda, R., Abel, S. J., Brown, W. O. J., Williams, S.,
Fochesatto, J., Brioude, J., and Bower, K. N.: The VAMOS Ocean-Cloud-Atmosphere-Land Study Regional Experiment (VOCALS-REx): goals, platforms, and field operations,
Atmos. Chem. Phys., 11, 627–654, doi:10.5194/acp-11-6272011, 2011.
Wyant, M. C., Wood, R., Bretherton, C. S., Mechoso, C. R.,
Bacmeister, J., Balmaseda, M. A., Barrett, B., Codron, F., Earnshaw, P., Fast, J., Hannay, C., Kaiser, J. W., Kitagawa, H.,
Klein, S. A., Köhler, M., Manganello, J., Pan, H.-L., Sun, F.,
Wang, S., and Wang, Y.: The PreVOCA experiment: modeling
the lower troposphere in the Southeast Pacific, Atmos. Chem.
Phys., 10, 4757–4774, doi:10.5194/acp-10-4757-2010, 2010.
Yang, M., Blomquist, B. W., and Huebert, B. J.: Constraining the
concentration of the hydroxyl radical in a stratocumulus-topped
marine boundary layer from sea-to-air eddy covariance flux measurements of dimethylsulfide, Atmos. Chem. Phys., 9, 9225–
9236, doi:10.5194/acp-9-9225-2009, 2009.
Yang, M., Huebert, B. J., Blomquist, B. W., Howell, S. G.,
Shank, L. M., McNaughton, C. S., Clarke, A. D., Hawkins, L. N.,
Russell, L. M., Covert, D. S., Coffman, D. J., Bates, T. S.,
Quinn, P. K., Zagorac, N., Bandy, A. R., de Szoeke, S. P.,
Zuidema, P. D., Tucker, S. C., Brewer, W. A., Benedict, K. B.,
and Collett, J. L.: Atmospheric sulfur cycling in the southeastern
Pacific longitudinal distribution, vertical profile, and diel variability observed during VOCALS-REx, Atmos. Chem. Phys., 11,
5079–5097, doi:10.5194/acp-11-5079-2011, 2011.
Yang, Q., W. I. Gustafson Jr., Fast, J. D., Wang, H., Easter, R. C.,
Morrison, H., Lee, Y.-N., Chapman, E. G., Spak, S. N., and
Mena-Carrasco, M. A.: Assessing regional scale predictions
of aerosols, marine stratocumulus, and their interactions during VOCALS-REx using WRF-Chem, Atmos. Chem. Phys., 11,
11951–11975, doi:10.5194/acp-11-11951-2011, 2011.
Zaveri, R. A. and Peters, L. K.: A new lumped structure photochemical mechanism for large-scale applications, J. Geophys. Res.,
104, 30387–30415, 1999.
Zaveri, R. A., Easter, R. C., Fast, J. D., and Peters, L. K.: Model
for simulating aerosol interactions and chemistry (MOSAIC), J.
Geophys. Res., 113, D13204, doi:10.1029/2007JD008782, 2008.
Zhu, P., Albrecht, B. A., Ghate, V. P., and Zhu, Z.: Multiplescale simulations of stratocumulus clouds, J. Geophys. Res., 115,
D23201, doi:10.1029/2010JD014400, 2010.
www.atmos-chem-phys.net/12/3045/2012/